AMD Radeon RX 580 vs NVIDIA GeForce GTX 1660 SUPER Comparison

# NVIDIA GeForce GTX 1660 SUPER vs AMD Radeon RX 580

The NVIDIA GeForce GTX 1660 SUPER and AMD Radeon RX 580 are both end-of-life graphics cards that shipped at the same launch MSRP of 229 USD, yet their benchmark profiles diverge sharply. Across the ten head-to-head tests in the data, the GTX 1660 SUPER wins every single matchup, with deltas ranging from a modest 4.7% in 2D work to a dominant 73.3% in DirectX 11. The RX 580, however, counters with a larger 8 GB memory buffer and a higher raw FP32 compute rating, making the comparison less one-dimensional than the win count suggests. The average benchmark scores are nearly identical — 12986 for the GTX 1660 SUPER versus 12928 for the RX 580 — placing both cards at the 53rd percentile of all GPUs, but that aggregate hides the per-test distribution where the NVIDIA card consistently pulls ahead.

Where Each One Wins

The GTX 1660 SUPER is the clear winner across every measured workload category, but the magnitude of its advantage varies by workload type. In API-specific rasterization tests, the NVIDIA card dominates: the DirectX 11 test shows a 73.3% delta, the DirectX 9 test shows 52.4%, and even the modern DirectX 12 test shows 16.3%. This pattern indicates that the Turing architecture's driver optimization and geometry handling provide a substantial edge in legacy and current-generation DirectX titles alike. The compute-oriented workloads also favor NVIDIA heavily — Geekbench OpenCL shows a 40.1% advantage, and Passmark GPU Compute shows a 45.5% advantage — suggesting that the GTX 1660 SUPER is the better choice for GPU-accelerated productivity tasks, not just gaming.

The RX 580, despite losing all ten head-to-head tests, retains relevance in scenarios where memory capacity matters more than raw throughput. Its 8 GB GDDR5 frame buffer doubles the GTX 1660 SUPER's 6 GB, which is advantageous for texture-heavy workloads at high resolutions or for compute tasks that require large working sets. The RX 580 also has a higher FP32 rating at 6.175 TFLOPS versus 5.027 TFLOPS, and a higher texture fill rate at 193.0 GTexel/s versus 157.1 GTexel/s, meaning it can theoretically sustain more pixel-fill work in shader-limited scenarios. However, benchmark results do not reflect these theoretical advantages — the GTX 1660 SUPER wins the Passmark G3D test by 44.1% — which suggests the NVIDIA card's architectural efficiency overcomes the raw spec deficit.

Architecture Differences

The two cards represent fundamentally different design philosophies from their respective manufacturers. The GTX 1660 SUPER uses the TU116 chip built on TSMC's 12 nm process, employing the Turing architecture with 6,600 million transistors on a 284 mm² die. The RX 580 uses the Polaris 20 chip on GlobalFoundries' 14 nm process, employing the GCN 4.0 architecture with 5,700 million transistors on a 232 mm² die. The NVIDIA chip packs more transistors into a larger die, yielding a transistor density of 23.2M / mm², while AMD's chip achieves a slightly higher density of 24.6M / mm² despite the older process node.

The compute resource allocation differs dramatically. The RX 580 has 2304 shading units, 144 texture mapping units, and 32 ROPs, whereas the GTX 1660 SUPER has 1408 shading units, 88 TMUs, and 48 ROPs. AMD's GCN design relies on massive shader counts, but NVIDIA's Turing architecture achieves higher per-shader efficiency, as evidenced by the FP16 throughput: the GTX 1660 SUPER delivers 10.05 TFLOPS FP16 (2:1 ratio), while the RX 580 delivers 6.175 TFLOPS FP16 (1:1 ratio). The memory subsystems also diverge: the GTX 1660 SUPER uses 6 GB of GDDR6 on a 192-bit bus for 336.0 GB/s bandwidth, while the RX 580 uses 8 GB of GDDR5 on a 256-bit bus for 256.0 GB/s bandwidth. The NVIDIA card's higher bandwidth and newer memory technology compensate for its smaller capacity.

Feature support differs in API versions. The GTX 1660 SUPER supports DirectX 12 (12_1) and Vulkan 1.4, while the RX 580 supports DirectX 12 (12_0) and Vulkan 1.3. Both support OpenGL 4.6. Neither card has ray tracing or tensor cores, as those are reserved for higher-tier Turing parts. The power profiles are starkly different: the GTX 1660 SUPER has a 125 W TDP with a 300 W suggested PSU, while the RX 580 has a 185 W TDP with a 450 W suggested PSU, making the NVIDIA card significantly more power-efficient per unit of performance.

Head-to-Head Benchmarks

The largest single-test victory for the GTX 1660 SUPER comes in the Passmark DirectX 11 test, where it scores 104 against the RX 580's 60, a 73.3% delta. This result is striking because DirectX 11 remains a widely used API in esports and older titles, meaning the NVIDIA card offers a substantial performance uplift in a broad swath of real-world gaming scenarios. The DirectX 9 test shows a similar pattern — 189 versus 124, a 52.4% delta — confirming that the NVIDIA card's advantage extends to legacy APIs as well.

In the 3DMark Steel Nomad DX12 test, the GTX 1660 SUPER scores 1278 against the RX 580's 1005, a 27.2% delta. This is the most modern workload in the suite, and while the NVIDIA advantage narrows compared to older APIs, it remains decisive. The Geekbench tests reinforce this trend: OpenCL shows 52490 versus 37453 (40.1% delta), and Vulkan shows 57102 versus 45173 (26.4% delta). The Vulkan gap is notable because Vulkan is a low-overhead API that tends to favor AMD's GCN architecture, yet the NVIDIA card still wins by over a quarter.

The Passmark G3D test, which aggregates several DirectX workloads, shows 12699 versus 8813, a 44.1% delta. The GPU Compute test shows 5076 versus 3488, a 45.5% delta. Even the 2D test, where the cards are closest, shows the GTX 1660 SUPER ahead at 805 versus 769, a 4.7% delta. The smallest deltas are in the DirectX 12 test (16.3%) and the G2D test (4.7%), while the largest are in DirectX 11 (73.3%) and DirectX 9 (52.4%). This distribution suggests that the NVIDIA card's advantage grows as the API becomes more driver-dependent and shrinks in the most modern, low-level APIs where raw hardware resources matter more.

FAQ

Q: Which card has a higher average benchmark score?

A: The NVIDIA GeForce GTX 1660 SUPER has an average benchmark score of 12986, while the AMD Radeon RX 580 has an average of 12928, a difference of 0.4% in favor of the NVIDIA card.

Q: Does the RX 580 win any benchmark test against the GTX 1660 SUPER?

A: No. Across all ten head-to-head benchmarks, the GTX 1660 SUPER wins every test, with deltas ranging from 4.7% in Passmark G2D to 73.3% in Passmark DirectX 11.

Q: How do the memory configurations differ between the two cards?

A: The GTX 1660 SUPER has 6 GB of GDDR6 on a 192-bit bus with 336.0 GB/s bandwidth, while the RX 580 has 8 GB of GDDR5 on a 256-bit bus with 256.0 GB/s bandwidth.

Q: Which card has a higher FP32 compute rating?

A: The RX 580 has a higher FP32 rating at 6.175 TFLOPS, compared to the GTX 1660 SUPER's 5.027 TFLOPS. However, the GTX 1660 SUPER wins the Passmark GPU Compute test by 45.5%.

Q: Are both cards the same price at launch?

A: Yes, both cards had a launch MSRP of 229 USD.

Q: What are the TDP and PSU requirements for each card?

A: The GTX 1660 SUPER has a 125 W TDP and requires a 300 W suggested PSU, while the RX 580 has a 185 W TDP and requires a 450 W suggested PSU.

Specification Differences

| Specification | NVIDIA GeForce GTX 1660 SUPER | AMD Radeon RX 580 |

|---|---|---|

| Architecture | Turing | GCN 4.0 |

| Process Node | 12 nm | 14 nm |

| Foundry | TSMC | GlobalFoundries |

| Transistors | 6,600 million | 5,700 million |

| Die Size | 284 mm² | 232 mm² |

| Transistor Density | 23.2M / mm² | 24.6M / mm² |

| Base Clock | 1530 MHz | 1257 MHz |

| Boost Clock | 1785 MHz | 1340 MHz |

| Memory Clock | 1750 MHz (14 Gbps effective) | 2000 MHz (8 Gbps effective) |

| Memory Size | 6 GB | 8 GB |

| Memory Type | GDDR6 | GDDR5 |

| Memory Bus Width | 192 bit | 256 bit |

| Memory Bandwidth | 336.0 GB/s | 256.0 GB/s |

| Shading Units | 1408 | 2304 |

| TMUs | 88 | 144 |

| ROPs | 48 | 32 |

| Pixel Rate | 85.68 GPixel/s | 42.88 GPixel/s |

| Texture Rate | 157.1 GTexel/s | 193.0 GTexel/s |

| FP32 | 5.027 TFLOPS | 6.175 TFLOPS |

| FP16 | 10.05 TFLOPS (2:1) | 6.175 TFLOPS (1:1) |

| TDP | 125 W | 185 W |

| Suggested PSU | 300 W | 450 W |

| Display Outputs | 1x DVI, 1x HDMI 2.0, 1x DisplayPort 1.4a | 1x HDMI 2.0b, 3x DisplayPort 1.4a |

| DirectX Support | 12 (12_1) | 12 (12_0) |

| Vulkan Support | 1.4 | 1.3 |

| Length | 229 mm (9 inches) | 241 mm (9.5 inches) |

| Release Date | 2019-10-28 | 2017-04-17 |